[DO NOT MERGE] Add Tutorials Audit Framework#3815
Conversation
Automated audit framework for PyTorch tutorial content health. Scheduled monthly via GitHub Actions (Stage 1: deterministic script-based audits) with optional Claude Code semantic analysis (Stage 2). Audit passes: - Build log warnings: extract DeprecationWarning/FutureWarning from CI logs - Changelog diff: cross-reference PyTorch release notes against tutorials - Orphaned tutorials: detect invisible tutorials, broken cards, NOT_RUN accountability - Security patterns: torch.load without weights_only, eval(), non-HTTPS URLs - Staleness check: tutorials-review-data.json freshness analysis - Dependency health: import vs requirements.txt mismatches - Template compliance: author attribution, grid cards, conclusion sections - Index consistency: tag typos, missing thumbnails, redirect chains - Build health: metadata.json coverage, shard imbalance, NOT_RUN growth Security: - Content sanitization (HTML comments, @mentions, script tags, truncation) - Claude skill with 6 mandatory guardrails (no PR actions, no file modifications) - Safe AST-based parsing of redirects.py (no exec()) - Streaming zip download for build logs (no full memory load) - Action versions pinned to SHA hashes Config-driven for cross-repo adoption. Only config.yml differs per repo. Trend tracking via previous closed audit issue (no contents:write needed). 68 pytest tests covering security boundary and all audit passes.
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/tutorials/3815
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 37bae8c with merge base cc4874c ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
sekyondaMeta
added
skip-link-check
|
Hi @sekyondaMeta! Thank you for your pull request. We require contributors to sign our Contributor License Agreement, and yours needs attention. You currently have a record in our system, but the CLA is no longer valid, and will need to be resubmitted. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks! |
1 participant
Automated audit framework for PyTorch tutorial content health. Scheduled monthly via GitHub Actions (Stage 1: deterministic script-based audits) with optional Claude Code semantic analysis (Stage 2).
Audit passes:
Security:
Config-driven for cross-repo adoption. Only config.yml differs per repo. Trend tracking via previous closed audit issue (no contents:write needed). 68 pytest tests covering security boundary and all audit passes.